Download First Principles of a Gas Discharge Tube (GDT) Primary

First Principles of a Gas Discharge Tube
(GDT) Primary Protector
Rev. 2
Circuit Protection Solutions
Written By Tim Ardley B.Sc (Hons)
Sr. Telecom Circuit Protection Field Applications Engineer
Table of Contents
1 Introduction
1
2 Bourns® discrete and 5-pin primary protection solutions
2
3 General operation of a GDT
4
4 GDT electrical parameters and their relationships
5
4.1 2026 series DCBD voltage and ramp impulse breakdown voltage
5
4.2 2026 series DCBD voltage and surge impulse voltage
6
4.3 2026-35 DCBD voltage and 1000 V/µs impulse voltage
7
4.4 2026-35 DCBD voltage and 10/1000 µs surge impulse voltage
7
4.5 2026-35 1000 V/µs impulse voltage and 10/1000 µs surge impulse voltage
8
4.6 Section summary and recommendations
8
5 Dynamic performance with telecom surges 9
10
5.1 Switching performance
5.1.1 Peak voltage switching time
10
5.1.2 Arc region transition time
11
5.2 Surge performance
12
5.2.1 2026 series switching time and surge impulse voltage with 10/1000 µs
12
5.2.2 2026-35 switching time and surge impulse voltage with 10/1000 µs
13
5.2.3 2026-35 switching time and surge impulse voltage with 2/10 µs
14
5.2.4 2026-35 switching time and surge impulse voltage with 8/20 µs
16
5.3 Surge voltage under extreme conditions
16
5.4 Arc voltage
18
5.4.1 Arc voltage under AC conditions
18
5.4.2 Arc voltage under surge
19
i
Page
5.5 Section summary and recommendations
20
6 Life cycle of a GDT
21
6.1 Longevity and end of life
22
6.2 The dark effect
23
6.3 The spark effect
24
6.4 GDT resistance below its breakdown voltage
25
6.5 Section summary and recommendation
26
7 Summary
27
8 Appendix Key differences between a GDT and TISP® thyristors
28
9 References
29
www.bourns.com
Table of Figures
Page
Figure 1 GDT switching characteristic
4
Figure 2 Ramp impulse voltage vs DCBD voltage options 5
Figure 3 2026 series surge impulse voltage during 1 kV, 100 A 10/1000 µs
6
Figure 4 DCBD voltage vs 1000 V/µs ramp impulse voltage
7
Figure 5 Variation of 2026-35 with 100 A 10/1000 µs surge 9
Figure 6 2026-35 GDT switching characteristic with 1 kV, 100 A 10/1000 µs surge
10
Figure 7 Reducing the impulse sparkover value of the GDT
11
Figure 8 Arc region transition time
12
Figure 9 2026-xx series switching time with 10/1000 µs surge
13
Figure 10 2026-35 switching time with 10/1000 µs surge
14
Figure 11 2026-35 switching time with 2/10 µs surge
15
Figure 12 Surge voltage versus generator short circuit current
17
Figure 13 Arc voltage variation during a 10/1000 µs surge
19
Figure 14 2035-20 DCBD voltage variation during repetitive surges
21
Figure 15 2026-35 life test during 1000 surges of 500 A 10/1000 µs
22
Figure 16 2026-35 surge impulse vs DCBD with 500 A 10/1000 µs
23
Figure 17 2026-35 dark effect
24
Figure 18 Spark effect during repetitive 500 A 10/1000 µs surges
25
Figure 19 Average resistance variation with 500 A 10/1000 µs
26
ii
1
Introduction
T
he earliest reference to experiments involving lightning was Benjamin Franklin’s kite experiments in
1752. However, it was in 1859 that Yale University described the use of the first spark gap protection
solution. The document uses Franklin’s experiment to harness the energy of a lightning storm. The kite
was tied to an insulated post where the metallic string going to the kite was connected to a brass ball on
the post. A second brass ball was grounded with a 2-inch gap between it and the isolated ball. If the energy
on the kite string became too intense, an arc would occur between the two brass balls to conduct the
energy to ground.
It was Humphrey Davy who investigated the basic spark gap and the nature of the arc between the
conductors. He discovered that with enough potential difference between the two conductors, the air or
the gas between the electrodes would ionize to provide a conductive medium for the current that would
produce the arc. The traditional brass ball spark gaps evolved into carbon blocks that still use the local
atmosphere to control the point of the arc. Spark gap technology has evolved where the atmosphere or
gas between the two electrodes was confined in a container so that the external atmosphere did not affect
its arcing performance. The gas discharge tube (GDT) as it has become known has evolved to provide
reliable and effective protection solutions during lightning storms and other electrical disturbances. GDTs
have been replacing carbon blocks in central office and remote access equipment installations due to
their robustness and better electrical performance in the traditional POTS (Plain Old Telephone Service)
type applications. With the recent explosion of broadband (digital) applications over copper, the GDT
is becoming the market leader, even over solid state (semiconductor solutions) due to their ultra-low
capacitance along with low insertion and return loss performances.
Gas discharge tubes are extensively used for the protection of telecommunications plant and equipment
against overvoltages caused by lightning, power switching and fault conditions. This document covers:
1
•
Bourns part numbering for discrete components and 5-pin modules and the general operation
of a GDT.
•
Investigating how the key electrical parameters and their relationships relate to each other with ramp
and surge waveforms.
•
Defining the expected ramp impulse and surge impulse of a GDT.
•
The life cycle of a GDT under surge and how its parameters change over its life.
www.bourns.com
Bourns® Discrete Components &
5-Pin Primary Protection Solutions
2
T
he Bourns® discrete components GDT family are available in both 2-electrode and 3-electrode options.
The 3-electrode, common chamber GDTs protect both the TIP and RING (outside conductors) to
ground (middle conductor). The series is available with a variety of lead configurations. The 3-electrode
options are also available with a mechanical switch grade fail-short mechanism. The Bourns® 3-electrode
common chamber GDT has the benefit of ionizing the second chamber simultaneously to aid both the TIP
and RING lines to switch into a virtual short circuit (arc region) at the same time. This is especially important
with transformer coupled interfaces where a large differential voltage across the line-side (primary) of the
transformer will transfer a higher secondary voltage. Using two discrete GDTs to provide the same function
will cause a scenario where one GDT will operate before the other, thus one side of the transformer will be at
ground potential while the other side is at or below the GDT’s impulse sparkover voltage. The Bourns® GDT
series are:
3-electrode discrete through-hole options
The 2026-xx-XX series is a through-hole solution available with DC breakdown (DCBD)
voltages from 75 V to 600 V and is rated for a single surge of 40 kA 8/20 µs and >10 operations of
20 kA 8/20 µs. The 2026-xx-XX series is also available with the mechanical switch grade fail-short
mechanism in a variety of lead configurations and is available in RoHS compliant versions. The
2026 series is about 11.5 mm (length) x 8 mm (diameter) without the fail-short mechanism, or with
a diameter of less than 10 mm with the fail-short mechanism.
The 2026-xx-XX-MSP series has precision-matched metal oxide varistors (MOVs) in parallel, which are
connected across each of the GDT chambers to ground. Adding the MOVs reduces the impulse breakdown
voltage under fast rising voltage transients. The series is rated as well for a single surge of 40 kA 8/20 µs and
20 kA 8/20 µs with more than 10 operations. The Bourns® MSP® series also has the mechanical fail-short
mechanism included and has a length of about 14 mm and a height of less than 10 mm.
The 2036-xx-XX series is the miniature through-hole option and is rated for a single surge of 20 kA
8/20 µs and greater than 10 operations of 10 kA 8/20 µs. The 2036-xx-XX series has dimensions of
approximately 7.5 mm (length) x 5 mm (diameter) and is available with DCBD voltages from
75 V to 600 V. The series is also available with the mechanical switch grade fail-short mechanism
that adds approximately one mm to the diameter.
3-conductor discrete surface mount options
The 2036-xx-SM series is the miniature option with DCBD voltages from 75 V to 600 V. The series
is rated as the standard 2036 series and has dimensions of approximately 7.5 mm (length) x 5.2 mm
(diameter). The 2036-xx-SM has approximately twice the DC sparkover voltage rating between the
TIP to RING conductors compared to its TIP/RING to ground conductors.
The 2038-xx-SM miniature series builds on the 2036-xx-SM but provides a symmetrical DC sparkover
voltage rating between any two of the three conductors. The 2038-xx series will be available with DCBD
voltages from 150 V to 1100 V and is rated for a single 10 kA 8/20 µs surge and a minimum of 10 operations
of 5 kA 8/20 µs. The dimensions are identical to the 2036 series of 7.5 mm (length) x 5.2 mm (diameter).
2-electrode discrete through-hole options
The 2027-xx-XX series is a through-hole package solution with options of formed and straight leads. The
series is rated for a single surge of 20 kA 8/20 µs and >10 operations of 10 kA 8/20 µs. The mechanical
dimensions are 6 mm (length) and 8 mm (diameter). The DCBD voltages are available from 90 V to 600 V.
2
2
The 2037-xx-XX series is a through-hole package solution that is rated for 10 kA 8/20 µs with
dimensions of 5 mm (length) x 5 mm (diameter). The DCBD voltage range is from 90 V to 600 V.
The 2035-xx series is a through-hole package solution that is rated as the 2037 series, but with a
slightly shorter body length of about 4 mm.
2-electrode discrete surface mount options
The 2027-xx-SM series is a surface mount solution of the standard 2027-xx-XX series with the same
ratings. The length is approximately 6.4 mm and the diameter 8 mm.
The 2035-xx-SM series is a surface mount solution of the 2035 series with identical ratings. The
length is approximately 4.4 mm and the diameter is 5 mm. The 2035-xx-SM has been designed for
robustness and longevity, with a voltage option from 90 V to 600 V.
The 2039-xx-SM series has DCBD voltage options from 470 V to 1100 V. This series extends the DCBD range
of the miniature GDT series and is rated for 5 kA 8/20 µs. The dimensions are approximately 4.4 mm length
and 5 mm diameter.
The 2055-xx-SM series provides a DCBD voltage range of 230 V to 420 V and is rated for 2.5 kA 8/20 µs. The
dimensions are 4.2 mm length and 6.2 mm diameter. The 2055-xx-SM has been designed for customer premise
equipment (CPE) where high surge current ratings are not as important as price.
Plug-in 5-pin primary protection selection
The worldwide telecommunications industry has come to rely on Bourns® surge protection devices to protect
sensitive communications equipment from electrical surges. Bourns offers a full range of surge protector
modules for central office, outside plant and customer premise. The customer premise modules have been
designed to fit into the NID (Network Interface Device, which Bourns also provides), mounted on the side of
the property. The 5-pin module is used in central office and remote access locations, where two basic series of
our 5-pin protectors can be considered:
The 2410 series modules - The patented Bourns® MSPs® (Multi-Stage Protectors) are based on the
discrete 2026-xx-XX-MSP series of 3-element GDTs and provide superior thermal protection with the
patented Bourns® “switch grade fail-short” mechanism. The 2410 series of 5-pin modules have been
designed to improve transmission performance and reach in broadband circuits based on balance and
a lower capacitance of less than 15 pF (the capacitance is due to the MSPs® hybrid design).
The 2420 series modules - The Bourns® 2420 series is based on the discrete 2026-xx-XX series, common
chamber 3-element GDTs and the switch grade fail-short device. It is UL listed according to the UL Seal-Test
program and recognized to be used without an air-back-up Gap (BUG) device. The BUG-less feature results
in longer trouble-free service life by eliminating the most common cause of premature failure of the protector
assembly. The 2420 also has ultra-low capacitance of less than 5 pF to make them suitable for the emerging
high-speed broadband applications.
The 2410 and 2420 series are UL listed and analyzed according to Telcordia GR 974 and SR 5165 requirements.
Both products meet CAT 5 requirements. The 2410 and 2420 series feature gold plated solid brass pins,
optional test access points and optional sealed unit construction. Tin plated pins are available on request.
Note: All Bourns® telecom GDTs and protector units are UL listed according to the UL Seal Test program and
Bourns® 2410-MSP and 2420 BUG-less series of 5-pin primary protectors are analyzed by Telcordia according
to the requirements of GR-974-CORE and SR 5165, reference Telcordia Analysis Report DA 1547.
3
www.bourns.com
General Operation of a GDT
3
W
Voltage
hen a voltage disturbance reaches the GDT sparkover value, the GDT will switch into a virtual short,
known as the arc mode. In the arc mode, the GDT practically shorts the line, diverting the surge
current through the GDT to ground and removes the voltage surge from the plant and equipment. At
normal operating voltages below the GDT rated DC sparkover voltage (measured at a rate of rise between
100 to 2000 V/s), the GDT remains in a high impedance off-state condition. All of our GDTs are rated at
a minimum insulation resistance of 109 Ω at 200 V, 100 V bias or 50 V bias for the different breakdown
values. With an increase in voltage across its conductors, the GDT will enter into its glow voltage region.
The glow region is where the gas in the tube starts to ionize due to the charge developed across it. During
the glow region the increase of current flow will create an avalanche effect in gas ionization that will
transition the GDT into a virtually short circuit mode
and current (dependent on the impedance of the
Sparkover Voltage
voltage source) will pass between the two conductors.
The voltage developed across the GDT with a short
circuit condition is called the “Arc Voltage” VARC.
The transition time between the glow and arc region
is dependent on the available current of the impulse,
Arc Region
Glow Region
the distance and shape of the electrodes, the gas
composition, gas pressure and the proprietary emission
coatings. Bourns active emission coatings allow the
tubes to transition into the arc mode at currents much
lower than 500 mA.
Arc Voltage
Time
Figure 1 – Ramp impulse voltage vs DCBD
voltage options
The arc voltage is the voltage rating that is specified
as <10 V @ 1 A for Bourns® GDTs. The GDT will
switch back or reset into a high impedance state once
there is not enough energy to keep the device in
the arc condition. This is sometimes also called the
“Extinguishing Voltage, Holdover Voltage or Impulse
Reset Voltage”. Figure 1 shows the basic switching
characteristic of a GDT.
Due to the typical arc voltage of less than 10 V, a GDT is ideal for protecting against high-energy impulses
and AC power cross conditions. Under AC, the power dissipated (VARC x IGEN) in the device needs to
be carefully monitored. The switch grade fail-short mechanism is available for the 2026/2036 series and
protects against thermal overload under AC conditions. At elevated GDT temperatures, a spring-loaded
clip will operate like a switch to short the TIP/RING conductors to ground. The 2026-xx-xx-F series is
recommended where high AC or long duration AC conditions exist or to meet Telcodia GR-974-CORE
requirements. The switch grade fail-short mechanism cannot be used to meet Telcordia GR-1089-CORE
requirements as there are first level AC tests where the equipment must work correctly after the test. It is
also not a good practice to hold a GDT in its glow region as this will significantly reduce the life expectancy
of the device. In this condition, significant heat can be developed on the electrodes that can damage the
special emission coatings and cause premature failure of the tube. Using a variable AC source such as a
curve tracer or equivalent to vary the voltage and power across at its DCBD voltage should be avoided. It
is highly unlikely that a condition could exist in the field which would maintain a GDT in the glow mode.
Bourns uses an active proprietary coating that provides a very low glow to arc transition current, helping to
reduce or eliminate these issues.
Note: The sparkover voltage shown in Figure 1 can be generated by either a ramp voltage defined as V/s or
V/µs or a peak surge voltage defined as TRISE/TFALL in microseconds. The sparkover voltage from a ramp
voltage waveform is referenced as the “ramp impulse voltage” where the sparkover voltage caused by a surge
waveform is a “surge impulse voltage”.
4
4
GDT Electrical Parameters and Their Relationships
T
his section investigates the relationship between the electrical specifications and how they relate to
the dynamic performance such as disturbances from lightning and AC. Understanding the primary
protector dynamic performance defines the expected performance under surge waveforms and aids the
selection of the secondary protection when assembled in modules or in close proximity of each other.
As an example, for the Bourns® GDT families we use the 2026 Series DCBD voltage and ramp impulse
breakdown voltage in the following discussion.
4.1
2026 Series DCBD Voltage and Ramp Impulse Breakdown Voltage
All the series of Bourns® GDTs have a wide variety of voltage options where the family’s DC breakover
voltage (voltage ramp rate of 100 V/s) were plotted against the ramp impulse voltage (voltage ramp rates
of 100 V/µs and 1000 V/µs) specified in the data sheet. Figure 2 shows the relationship between these
two parameters. It can be seen that the Bourns® GDT impulse rating is directly related to the DC series
breakdown voltage. The ratio of the DCBD voltage to ramp impulse voltage performance for each individual
voltage family is approximately 1.4:1 for 100 V/µs and 1.6:1 for 1000 V/µs. To calculate the typical impulse
performance with a DC series breakdown voltage (VDCBD) rating, the formula below can be used:
VIMPULSE @ 100 V/µs = (1.4 x VDCBD) + 140
VIMPULSE @ 1000 V/µs = (1.6 x VDCBD) + 300
2026 Series Impulse Performance
1400
Impulse Voltage
1200
1000
800
600
400
200
0
0
100
200
300
400
DC Breakdown Voltage
500
600
1000 V/µs
700
100 V/µs
Figure 2 – Ramp impulse voltage vs DCBD voltage options
Figure 2 shows that for instance the 2026-07 has a 75 VDCBD rating that does not follow the same curve as
the rest of the voltage family in the series and the deviation gets worse as the ramp impulse voltage increases.
This is due to the low voltage options having different gas and pressure properties to achieve the low voltage
breakdown which is at the limitation of the gas discharge technology. Figure 2 also shows that the 75 V and
90 V tubes have a higher ramp impulse performance than the 150 V option. For limiting the maximum
voltage across a twisted pair of cables under fast rising transients, the 150 V option may be a better
design proposal.
5
www.bourns.com
4
Note: The impulse voltage is specified in the data sheet, which should be used to determine the maximum
voltage protection value or “let-through voltage”. The DCBD voltage value can easily be measured with the
Bourns® 4010 handheld tester, where the relationship between a single device measurement and its dynamic
performance is discussed in sections 4.3 to 4.5. The low end DCBD value is the voltage that should be used
to determine the minimum required withstand voltage to avoid interference of the GDT with the normal
system operating voltage.
2026 Series DCBD Voltage and Surge Impulse Voltage
The 2026 series was measured for DCBD voltage and then exposed to a surge with a 1 kV, 100 A 10/1000
µs. The surge impulse voltage was measured where figure 3 shows that there is a correlation between the
different voltage ranges in the 2026 series and the surge impulse voltage. The relationship also looks to be
directly proportional to the measured DCBD voltage rating of the device, which is covered in section 4.3.
The ratio of DCBD voltage to surge impulse voltage for each voltage family is approximately 1.12:1 for the
10/1000 type surge as shown by the slope of the approximation line in figure 3.
2026 Series 1 kV, 100 A 10/1000 µs
1000
900
800
700
Surge Voltage
4.2
600
500
400
300
200
100
0
0
100
200
300
400
500
600
700
DC Breakdown Voltage
Figure 3 – 2026 series surge impulse voltage during 1 kV, 100 A 10/1000 µs
Note: The 2026-07 (75 V) and 2026-09 (90 V) options show noticeable performance differences compared
to the rest of the voltage options in the family. Figure 3 shows that the 150 V 2026-15 would provide a better
protection solution under common surge impulse waveforms while providing a higher maximum working
voltage before the GDT operates. The 75 V and 90 V series are ideal for very slow rising impulses.
Comparing figures 2 and 3 shows that using a 1000 V/µs rating of a GDT to define a 10/1000 µs surge
rating is not providing an accurate representation. This difference is approximately 57 % of additional
safety margin. The 100 V/µs impulse sparkover rating is much closer to the expected surge values and still
provides 17 % of additional tolerance under 10/1000 µs. Using the 100 V/µs appears to be a better method
for dynamic comparison between the two test methods. This is especially important when designing to
Telcordia GR-1089-CORE, issue 4 which uses some of the GDT dynamic performances to help define the
secondary protection requirements. The assumption can be made that using the 1000 V/µs impulse rating
will also cover faster surge impulse transients. However, the fast rising transients such as the 1.2/50 µs
voltage wave shape (8/20 µs current) only yielded a 7 % increase in surge voltage with the same peak open
circuit voltage. This area is covered in more detail in section 5; dynamic performance with telecom surges.
6
4
4.3
2026-35 DCBD Voltage and 1000 V/µs Impulse Voltage
A GDT life tester was used to measure the actual DCBD voltage and ramp impulse sparkover voltage at
1000 V/µs for each GDT. The 2026-35 (350 V) series was tested to see if there is individual device
correlation (higher DCBD voltage results in higher impulse voltage) between the dynamic parameters.
Section 4.1 showed that there was correlation between these two parameters when looking at the voltage
selection options in the GDT family. Figure 4 shows that there was a distinct grouping of the measurements,
but there was no direct correlation between the DCBD measurement and impulse voltage measurement.
DC Breakdown vs 1000 V/µs Impulse Voltage
900
880
860
Impulse Voltage
840
820
800
780
760
740
720
700
300
320
340
360
380
400
DC Breakdown Voltage
Figure 4 – DCBD voltage vs 1000 V/µs ramp impulse voltage
Note: It was not possible to confirm the results by using an independent impulse tester.
4.4
2026-35 DCBD Voltage and 10/1000 µs Surge Impulse Voltage
A 10 A 10/1000 µs surge was used to limit any possible electrical variation in surge impulse voltage due to
repetitive surges. As with the 1000 V/µs test in section 4.3, there was a distinct grouping of the results, but
again, no direct correlation between the two electrical parameters could be seen. The surge current was
increased to 100 A, 300 A and 500 A 10/1000 µs to see if this would provide any direct correlation with the
DCBD voltage and surge impulse voltage values. However, no direct correlation could be found.
Note: To reconfirm this, an independent test was done using the Bourns® 4010 handheld tester to measure
the DCBD voltage and the Keytek E509 surge generator. The test results were very similar to conclude the
original measurements were correct.
7
www.bourns.com
4
4.5
2026-35 1000 V/µs Impulse Voltage and 10/1000 µs Surge Impulse Voltage
The data collected in sections 4.3 and 4.4 was used to see if there was any direct correlation between the
1000 V/µs ramp impulse voltage measurement and 10/1000 µs surge impulse voltage measurements. There
was a distinctive grouping of the measurements, but again there was no direct correlation between these
two measured electrical parameters. Surging the devices 100 times for each surge current (10 A to 500 A)
increased the standard deviation (scatter area), but no direct correlation was found between the two electrical
parameters. This increased scatter effect started the investigation of the life expectancy of the GDT, as the
DCBD voltage was changing under each surge and was very noticeable with the 300 A and 500 A tests. This is
covered in more detail in section 6, life cycle of a GDT.
4.6
Section Summary and Recommendations
There was correlation found between each voltage in the 2026-xx-XX series, but not between each individual
GDT’s measured DCBD value and its measured impulse or surge impulse voltage value. Therefore, finding the
surge impulse voltage can be calculated by using the series DC voltage selection options in the data sheet.
The surge impulse voltage and ramp impulse voltage is only defined by the voltage and its rate of rise across it†.
The DCBD voltage selection in a GDT series can be used to provide a good indication of the expected surge
impulse voltage during a 10/1000 µs surge as shown in figure 3.
Defining the typical impulse voltage and surge impulse voltage from the DCBD voltage option can be
approximated by:
VIMPULSE @ 100 V/µs = (1.4 x VDCBD) + 180
VIMPULSE @ 1000 V/µs = (1.6 x VDCBD) + 300
VSURGE (10/1000 µs) = (1.2 x VDCBD) + 170
Note: These formulas can be used for DCBD voltages of 150 V or more.
With fast (greater than 100 V/µs) impulse and telecom surges, the 150 VDCBD option can provide better
protection than the 75 V and 90 V options in the GDT series. The benefit of the low voltage options is with
slow rising voltages of 2000 V/ms or less such as AC mains voltages for example.
† The minimum telecom surge current tested was 10 A which is significantly above the glow to arc current specification of 500 mA.
There may be GDT dynamic performance considerations if the short circuit current is close to the arc transition current which was not
explored in this document.
8
5
Dynamic Performance with Telecom Surges
T
he GDT impulse sparkover specifications are characterized by an impulse voltage waveshape such as
100 V/µs or 1000 V/µs. This ramp voltage is not an accurate representation of the real world scenario.
Electrical disturbances entering into the building by twisted pair cable from a lightning strike are based
on rise and decay times since the lightning strike is charge (Q) related being dissipated in an impedance of
the line. Section 4 showed there was a relationship between the DCBD voltage and both the ramp impulse
voltage and surge impulse voltage. This highlighted a typical value from a one lot batch. Different test units
were selected from different batches of GDTs and surged on two separate 100 A 10/1000 µs generators.
Figure 5 shows the results plotted on the same curve shown in figure 3 and the 100 V/µs curve shown in
figure 2. Figure 5 shows that there are similar performances between the 100 V/µs impulse breakdown
voltage and the 100 A 10/1000 µs surge impulse voltage performance.
The formula on page 8 is a good approximation if the 100 V/µs ramp impulse voltage is not specified in the
data sheet for the GDT.
100 V/µs Ramp Voltage and 10/1000 µs Surge
1000
900
Surge Impulse Voltage
800
700
600
500
400
300
200
100
0
0
100
200
300
400
500
600
700
DC Breakdown Voltage
Figure 5 – Variation of 2026-35 with 100 A 10/1000 µs surge
Field studies have shown that the strike energy is similar to a 10/250 µs waveform for a positive stroke.
The primary protection telecom standards such as Telcordia GR-974-CORE also specify end of life mode
tests using a series of surge generators of 10/1000 µs, 10/250 µs and 8/20 µs type waveshapes. The dynamic
performance of the GDT with these telecom surge waveshapes are less known. As secondary protection
is often required to protect voltage sensitive equipment from the primary protector let-through, it is
important to know how the GDT performs under these types of impulses. Standards such as Telcordia
GR-1089-CORE for secondary protection are still based on worst case, carbon block type primary
protection technology to cover legacy equipment. GR-1089-CORE, issue 4 release allows the secondary
protection to be more dependent on the primary protector technology used in the field.
9
www.bourns.com
5
5.1
Switching Performance
The 2026-35 was surged with a 1 kV, 100 A 10/1000 µs. The current and voltage waveforms are shown in
figure 6. From the curve, it can be seen that the glow region is very short (cannot actually be seen) under
this condition. The two switching times to consider during the operation of the GDT are the peak voltage
breakdown and the transition into the arc region.
2026-35 Switching Characteristic
140
600
120
500
100
400
80
300
60
200
40
100
20
Current - Amps
Voltage - Volts
700
0
0
-20
-100
0
2
4
6
Time - µs
8
Voltage
10
Current
Figure 6 – 2026-35 GDT switching characteristic with 1 kV, 100 A 10/1000 µs surge
5.1.1
Peak Voltage Switching Time
This is the time measured from the start of the impulse applied to the GDT to when the GDT switches
into a virtual short circuit. Figure 7 shows the 2026-35 switching characteristic with a peak surge impulse
voltage of 588 V. The voltage curve rising edge is the rise time (10 µs to 1000 V) of the surge generator before
the GDT switches into a virtual short and started further investigation covered in section 5.2.1 to 5.2.3.
This indicates that the surge impulse voltage is directly related to the generator open circuit voltage and is
only voltage rise dependent. Therefore, if a GDT conducts at a lower peak voltage, it will provide a faster
switching time. Figure 7 shows the dynamic impact of a GDT operating at a 450 V surge impulse sparkover
compared to 550 V. The 450 V option will have a typical switch time of 3.4 µs where the 550 V will be 4.1 µs
(reduction of 0.7 µs with 100 V less let-through). As the rising edge (dV/dt) increases, the switching time
also reduces, since the impulse sparkover performance is normalized between the two. This benefit needs
to be considered when compared against the reduced life span of a faster acting GDT. For a GDT used as a
primary protector, the longevity and current withstand capability over the minor savings of faster switching
performances may be more important. A GDT primary protector providing a tighter protection window
to reduce the secondary requirement should be a serious consideration. However, without a significant
reduction in the impulse voltage specifications, like the solid-state TISP® thyristor, for example, the actual
impact is minimal to the secondary protection in the equipment. GR-1089-CORE, issue 4 specifies three
forms of primary protector that can be considered. The solid state (low voltage) needs to develop 400 V
across it to ensure operation, a GDT (medium voltage) is 600 V and a carbon block (high voltage) is 1000 V.
This provides clear differentiation between the different primary protection technologies.
10
5
140
600
120
500
100
400
80
300
60
200
40
100
20
0
Current - Amps
Voltage - Volts
1 kV 10/1000 µs GDT Switching Characteristic
700
0
-100
-20
1
2
3
4
5
6
Time - µs
Figure 7 – Reducing the impulse sparkover value of the GDT
5.1.2
Arc Region Transition Time
The representation of the arc region transition time is shown in figure 1 and is the time between the
impulse sparkover voltage to the arc voltage. Figure 7 shows the 2026-35 GDT has a very fast transition
time to 70 V where this transition time for the 2026-35 was in the region of 100 ns. The 2026-35 was surged
with different impulses to see if there was any difference in the arc transition time.
2/10 µs 2500 V, 500 A
8/20 µs 2500 V, 1000 A
10/1000 µs 2500 V, 1000 A
Surge
Voltage
(V)
Time
(ns)
Surge
Current
(I)
Surge
Voltage
(V)
Time
(ns)
Surge
Current
(I)
Surge
Voltage
(V)
Time
(ns)
Surge
Current
(I)
Min.
500
77.67
116
492
82.2
260
512
64
43.2
Max.
628
127.2
564
668
104.8
1210
616
76
110
Avg.
588
97
N/A
623
92
N/A
564
68
N/A
Table 1 – Transition time during a surge
The variation of transition time to surge current seen in figure 8 indicates the previous surge impacted the
transition time rather than the peak current or voltage. The impurities released from the tube’s internal
material can increase transition. The emission coatings used in the GDTs will attract (getter) the impurities
released which will make the next surge transition time revert to the initial values. This process will repeat
itself over the lifetime of the GDT. A summary of the results are shown in table 1.
11
www.bourns.com
5
2026-35 Arc Region Transition Time
1400
1200
Surge Current
1000
800
600
400
200
0
70
80
90
100
110
120
Time - ns
‘2/10
130
‘8/10
Figure 8 – Arc region transition time
5.2
Surge Performance
Section 4.2, figure 3 compares the different DCBD voltage selection options of the 2026 series under
10/1000 µs surge and indicates that there is correlation between the voltage family and the surge impulse
voltage. This section will investigate further how GDTs behave under different telecom surges.
5.2.1
2026 Series Switching Time and Surge Impulse Voltage with 10/1000 µs
The voltage options in the 2026 series were surged with 1 kV, 100 A 10/1000 µs waveform. The surge
impulse voltage and switching time was measured and plotted against the generator open circuit voltage
waveform. The results are shown in figure 9 where the performance of the different voltage options can be
seen. Figure 9 shows five groupings, while six voltage options were tested. The 90 V series had a slightly
higher impulse voltage compared to the 150 V series, and therefore, when plotted on the curve looked like
one distribution. Figure 9 also showed the surge impulse voltage or switching time tolerance varied with
voltage option and tracked the open circuit surge voltage characteristic. The lower breakdown voltage series
had a better standard deviation in switching times than the higher voltage options and the higher voltage
options had a better surge impulse voltage standard deviation for the same open circuit waveform.
A summary of the results from this test is shown in table 2.
2026-09
2026-15
2026-25
2026-35
2026-47
2026-60
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Avg.
352
1.48
330
1.43
435
1.93
551
2.73
668
3.65
832
6.08
Std.
7.14
0.04
10.58
0.03
9.63
0.06
21.92
0.10
21.77
0.25
17.22
0.76
Ratio
2.02
1.06
1.06
1.09
1.06
1.10
Table 2 – 10/1000 µs surge impulse voltage statistics
12
5
The ratio factor was calculated using the formula highlighted in section 3.6, where the
ratio = (VSURGE - 170) / VDCBD).
All the voltage options, except 90 V series had similar ratios. This showed that all the GDT voltage series
performed in the same manner and that their dynamic performance will be similar in ratio to their DCBD
value. The 350 V 2026-35 series is a popular option and will be used to further investigate the GDT during
surge conditions.
1 kV 10/1000 µs, 100 A Impulse Generator
1200
Surge Impulse Voltage - V
1000
800
600
400
200
0
0
5
10
15
20
Switching Time - µs
Figure 9 – 2026-xx series switching time with 10/1000 µs surge
5.2.2
2026-35 Switching Time and Surge Impulse Voltage with 10/1000 µs
The 2026-35 was surged with different generator open circuit voltages. The peak voltage from the impulse
generator was varied between 400 V to 1000 V where the peak surge impulse voltage and the open circuit
delay time was measured. The generator voltage was set to 400 V to ensure the GDT would still operate. At
400 V, the surge impulse voltage of the GDT was very consistent between each device. The wide variation
between each device occurred in the switching time. As the impulse generator voltage increased, the
switching time became more consistent, but the surge impulse voltage variation increased. Figure 10 shows
this variation between time and voltage. The generator open circuit voltage (blue curve) is also plotted on
figure 10 to provide a point of reference. Table 3 summarizes the test results.
13
www.bourns.com
5
10/1000 µs, 10 Ω Impulse Generator
1200
Peak Impulse Voltage - V
1000
800
600
400
200
0
-10
0
10
20
30
40
Time - µs
Figure 10 – 2026-35 switching time with 10/1000 µs surge
2026-35 with 10/1000 µs Impulse
Generator
Voltage
412 V
520 V
Impulse
(V)
Time
(µs)
Avg.
408
Std.
5.34
1048 V
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
28.75
473
6.80
8.79
10.38
596
2.73
1.22
14.94
.10
Table 3 – 2026-35 with different 10/1000 µs generator voltages
Note: Bourns® 2026-xx series will operate with a 10/1000 µs voltage which is 20 % above its DCBD voltage
specification. The typical surge impulse voltage can be found using the formula on page 8.
5.2.3
2026-35 Switching Time and Surge Impulse Voltage with 2/10 µs
The 2026-35 was surged using a 2/10 µs impulse where the results are shown in figure 11. The 2026-35 did
not switch into a virtual short circuit with a voltage of 400 V. The voltage was increased to 500 V to ensure
all the units operated. Figure 11 shows that the GDT was just operating at a peak generator voltage of 500 V,
which indicates this voltage is close to the minimum requirement for reliable GDT operation. Table 4 shows
the summary of the test.
14
5
2/10 µs, 10 Ω Impulse Generator
2000
1800
Peak Impulse Voltage - V
1600
1400
1200
1000
800
600
400
200
0
-1
0
1
2
3
4
Time - µs
Figure 11 – 2026-35 switching time with 2/10 µs surge
Note: Bourns® 2026-xx series will operate best with a 2/10 µs generator if the voltage is 50 % above its
DCBD voltage specification.
2026-35 with 2/10 µs Impulse
Generator
Voltage
512 V
600 V
864 V
1760 V
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Avg.
512
2.27
548
1.21
634
0.46
688
0.09
Std.
2.56
0.30
12.17
0.10
13.52
0.03
24.51
0.01
Table 4 - 2026-35 with different 2/10 µs generator voltages
When comparing the results of table 4 with table 3, the surge impulse voltage increased due to the faster
rising voltage of the 2/10 µs waveform while the switching times reduced. The surge impulse voltage
however only increased by 7 % between the two waveshapes. The same phenomenon was seen with the
10/1000 µs test where the variation between surge voltage and switching time changed depending on the
generator voltage. Section 4.2 highlighted that the ratio between the voltage options in the 2026 series
remained constant. Therefore, the original formula needs to be modified to take into consideration the
increased surge impulse rating. From table 4, the average surge impulse voltage was 688 V which changes
the formula highlighted on page 8 to approximate:
VSURGE = (1.2 x VDCBD) + 280
15
www.bourns.com
5
5.2.4
2026-35 Switching Time and Surge Impulse Voltage with 8/20 µs
The 2026-35 was surged with the 8/20 µs (1.2/50 µs voltage waveform) where the results are shown in
table 5. With an open circuit generator voltage of 400 V, only 70 % of the units tested switched into the arc
region. This indicated that the 350 V GDT series would need a minimum of 450 V to ensure it switched
into a virtual short circuit. This was comparable with the 2/10 µs results where the GDT was assured to
operate with a voltage of 500 V.
2026-35 with 8/20 µs Impulse
Generator
Voltage
500 V
610 V
1210 V
1740 V
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Impulse
(V)
Time
(µs)
Avg.
406
8.91
490
2.03
638
0.37
668
0.25
Std.
2.42
2.92
3.78
0.38
32.09
0.03
24.83
0.03
Table 5 - 2026-35 with different 8/20 µs generator voltages
Table 5 shows the same phenomenon of impulse and switching time variation of surge impulse voltage as
shown with the 2/10 µs and 10/1000 µs waveforms in tables 3 and 4. The results also highlight very similar
performances of the surge impulse voltage and switching times between the 2/10 µs waveform. It was
expected that the faster voltage rising edge of the 8/20 µs impulse would further increase the surge voltage
of the GDT, but this was not the case.
5.3
Surge Voltage Under Extreme Conditions
Section 5.2 showed the surge impulse voltage increased as the generator voltage increased. Extrapolating
the data points in figure 11 or 12 indicates that there will be a maximum surge impulse voltage. This
would define the worst case surge impulse voltage and therefore the maximum peak voltage seen by the
equipment. An 8/20 µs generator with a maximum voltage of 6 kV was used to investigate this further. The
generator series resistance was measured as 2.48 Ω and the current was monitored during the testing using
a Pearson Electronics current transformer.
Figure 12 shows that the surge impulse voltage developed across the GDT started to decline after 800 A
with an open circuit voltage of 1984 V (800 A x 2.48 Ω = 1984 V). The 2/10 µs, 10 Ω generator was also
used as a quick comparison. The peak surge impulse voltage occurred with an open generator voltage of
1700 V with a similar shaped curve.
16
5
2026-35 with 8/20 µs Surge Current
750
10000
650
1000
600
100
550
Switching Time (ns)
Surge Voltage (V)
700
500
10
450
200
700
1200
1700
2200
Surge Current (Amps)
2700
Surge
Voltage
3200
Switching
Time
Figure 12 – Surge voltage versus generator short circuit current
This shows that the worst stress surge voltage applied to the secondary protection or equipment is when the
GDT is subjected to a 2 kV 2/10 µs or 8/20 µs surge. Increasing the generator voltage past this point will
reduce the surge impulse voltage. The increased current will only be testing the ground return current path.
If the GDT data sheet is specified with a surge impulse voltage rating with a 2 kV, 2 Ω 8/20 µs surge, this
will help to ensure voltage coordination for the secondary protection on line cards, for example. This test
can also be useful for equipment being used in high lightning, remote access applications. GR-1089-CORE,
issue 4, section 4.7 covers lightning protection tests for equipment to be located in high exposure customer
premises or OSP facilities and uses a 10/250 µs generator up to 4 kV, 500 A capability. Further investigation
needs to be done with this waveform to see if 4 kV actually provides maximum stress on the equipment.
17
www.bourns.com
5
5.4
Arc Voltage
The arc voltage (VARC) is developed across the GDT during its “virtual short circuit” condition. This
parameter defines the power dissipation of the GDT during its protection mode. A low arc voltage is
desirable to keep power dissipation at a minimum, which in turn increases the life expectancy of the
tube due to a lower operating temperature. The miniature surface mount series GDTs are, in addition to
other applications, being considered for secondary protection in broadband solutions, but require other
considerations. For these types of applications, the overcurrent protectors need to be carefully selected to
ensure that the GDT is bypassed correctly. If this is not performed correctly, the GDT could be damaged
before the overcurrent protector operates under the AC power cross tests.
The arc voltage will also define how the primary and secondary protection solutions perform together.
The principle is to have an overcurrent protector between the primary and secondary overvoltage
protectors. The belief is that once the GDT operates, the virtual short of the GDT will redirect the
energy away from the secondary protection and the secondary will reset. It is common for a secondary
protector such as a TISP® thyristor to be used. If the overcurrent protector solution is a fuse, it has been
established that the fuse will need to clear before the GDT can operate if there is not enough impedance
between the two voltage protectors. Resistance to achieve voltage coordination to allow the protection
solution to reset is usually between 5 Ω to 50 Ω, but is dependent on the secondary protector withstand
current capability. The current through the TISP® thyristor has to fall below its holding current, IH (150
mA min. at an ambient temperature of 25 ºC) for it to reset back into a high impedance state. The TISP®
thyristor on-state voltage, VT is approximately 3 V (dependent on surge current and impulse shape), so
with a 50 Ω series resistance, the arc voltage must be lower than 10.5 V (0.15*50+3). Lower resistance
values compound the problem. Having a higher arc voltage GDT will increase the stress for the secondary
protection where a ceramic positive temperature coefficient (CPTC) thermistor, such as the Bourns®
CMF-SD50 series, would help in this instance.
5.4.1
Arc Voltage Under AC Conditions
The arc voltage is specified with a 600 VRMS, 1 A generator. To reconfirm the data sheet specifications,
the 2026-35 arc voltage was measured with 600 VRMS with 0.92 A and 3.4 A. The VARC did not change
between the two load current values where the typical arc voltage was just 8.1 V.
18
5
5.4.2
Arc Voltage Under Surge
The 2026-xx series was surged with a 1 kV, 100 A 10/1000 µs and the arc voltage measured. The
results highlighted a voltage variation from 11.78 V to 12.20 V between the different voltage options
in the series. This variation is too small to be further investigated in this report and therefore
concludes that there is no major variation between different voltage options in the GDT series. This
also showed that the arc voltage changed over the duration of the impulse. The arc voltage of the
2026-35 was re-measured at 1000 µs where the typical VARC was just 11 V as shown in figure 13. As
the delay time stayed constant during the original voltage measurements, comparisons could still be
made between the different voltages in the series.
Reducing the peak 100 A 10/1000 µs surge current by 50 % did not show any significant change in
the arc voltage as shown in figure 13. This shows that the arc voltage is not current dependent under
surge.
The 2026-35 was surged with a 2/10 µs impulse to see if the waveshape changed the arc voltage
characteristic. The arc voltage did not significantly vary between the 10/1000 µs and 2/10 µs impulse.
Arc Voltage During 10/1000 µs Surge
20
19
18
Voltage - V
17
16
15
14
13
12
11
10
0
0.5
1
Time (ms)
1.5
100 A Surge
2
50 A Surge
Figure 13 – Arc voltage variation during a 10/1000 µs surge
Note: It is recommended to define the time delay after the start of the surge impulse to allow
accurate comparisons of the arc voltage. For this application note, the VARC measurements were
taken at the decay value time of the surge waveform being used. Telecom surges specify the fall time
at 50 % of the peak value where the measurement for a 10/1000 µs surge was measured at 1000 µs
and 10 µs for a 2/10 µs.
19
www.bourns.com
5
5.5
Section Summary and Recommendations
The most suitable waveform for testing voltage coordination between the primary and secondary
protector is either 2/10 µs or 8/20 µs as this developed the highest surge impulse voltage. Telcordia
GR-1089-CORE, issue 4 specifies 1000 V/µs rating if the primary does not conform to GR-974CORE (specifies 600 V for a medium voltage protector). Using the 1000 V/µs specification for the
surge impulse voltage adds significant head room which makes the solution more robust, but at
the expense of the secondary protection being “over-engineered” for the task. The more realistic
specifications to use are the 100 V/µs specifications.
•
Using the VDCBD + 50 % to set the peak surge generator voltage will provide maximum stress to
the secondary protection with the GDT removed from the circuit. This also simplifies additional
testing to find the GDT threshold.
•
A 2 kV 8/20 µs generator voltage with a 2 Ω generator impedance produced the highest surge
impulse voltage of the 2026-35. Section 5.2 showed that there was no significant difference in
the surge impulse voltage between the 8/20 µs and 2/10 µs waveforms when using the same
generator open circuit voltage, so either waveform can be used to evaluate the protection circuit.
•
The impulse sparkover rating with a 100 V/µs is a good indication of the expected maximum
surge impulse voltage when considering the 10/1000 surge. The increase of surge impulse voltage
during a 8/20 µs or 2/10 µs impulse generator is minimal (7 %) which allows the 100 V/µs rating
to be a good approximation when designing to other telecom surge waveshapes.
•
To calculate the coordination voltage with a known series DC sparkover voltage option, use
VSURGE = (1.3 x VDCBD) + 280. This is for GDTs where the VDCBD is greater than 150 V.
GDT voltage selection options less than 150 V will provide higher surge impulse voltages and
will need another formula to calculate the expected surge impulse voltage.
•
The maximum stress applied to the secondary protection during a surge is just before the GDT
switches into a virtual short. This can be estimated as:
10/1000 µs surge is VDCBD + 15 %.
2/10 µs surge is VDCBD + 50 %.
Surging above these breakdown points of the GDT will reduce the energy seen by the
secondary protection.
•
Surging ports above 2 kV with telecom surges does not necessarily stress the secondary
equipment or test voltage coordination between the primary and secondary protector. The surge
impulse voltage can actually reduce with an increase in surge generator voltage. The increased
current with the higher generator voltage will test the durability of the GDT and the ground
return path of the GDT.
•
Using the 1000 V/µs rating of the GDT is not an accurate method for identifying peak surge
impulse voltage. The peak voltage seen was VDCBD + 113 % where the 1000 V/µs rating for the
2026-35 is VDCBD + 150 %. Using the 1000 V/µs rating of the GDT just increases the safety
tolerance, but can make the secondary protection circuitry more expensive than necessary.
•
Switching time is dependent on the rise time to impulse sparkover voltage rating of the GDT. A
fast rising edge waveform will provide a consistent switching time where a slow rising edge will
provide a consistent surge impulse voltage.
•
The arc region transition time to 70 V is less than 100 ns and is not significantly affected by the
surge waveform or impulse current. The arc voltage is also not affected by the impulse current or
waveshape. However, careful measurement is required.
20
6
Life Cycle of a GDT
T
he GDT does wear out due to particulates being dislodged from the electrodes during the arcing of the
tube. The impact of the arc across the tube is dependent on the energy strike, so the life of the GDT tube
will be dependent on the impulse applied to it. The life of the tube is expressed in a surge waveform and the
number of surges. The principle of solid state protectors such as the TISP® thyristor family having a wear
out mechanism is documented, but compared to the life span of a GDT, it can be considered not to have a
wear out mechanism. Thyristor measurements are also based on testing done using the “Step Stress” method
(increasing a generator voltage waveform sequentially while testing the electrical characteristics between
each surge for variation). The step stress test also needs the device under test to fail with a single surge.
The GDT proved difficult to run a step stress test as the DCBD voltage varied between each test and a
tolerance would need to be put in place to define a pass/fail criteria. The 40 kA 8/20 µs generator could
not guarantee that the unit would fail on the first test which is key in step stress testing. Semiconductor
protectors support lower impulse currents and have a tight window between an electrically good test
to a short circuit failure. These two parameters make the step stress principle work well in the electrical
characterization process for these types of devices, but not for the current family of Bourns® GDTs. The life
cycle experiments were therefore done using a “Repetitive Stress” test method (repetitively surging the GDT
with a fixed voltage/current impulse) where the DCBD was measured after each test.
Note: The 2035-20 (single element, 200 V) series was used in figure 14 due to it being half the impulse
rating of the 2026-xx series to accelerate the end of life point of the GDT.
2035-20 100 A 10/1000 µs
350
DC Breakdown Voltage - V
300
250
200
150
100
50
0
0
500
1000
1500
2000
2500
Number of Surges
Figure 14 – 2035-20 DCBD voltage variation during repetitive surges
The separation of the two DCBD voltage measurements (positive and negative polarity measurement)
shown in figure 14 is due to the particulates being attracted to a single conductor. The surge ionizing effect
charges the tube and therefore attracts the particulates to one end of the tube. This has the effect of changing
the electrical properties such as the DCBD.
Note: Alternating polarity (+/-) surging on each impulse test increases the longevity of the tube due to the
particulates being dislodged from each of the electrodes, rather than just one. The data captured in figures
14 and 15 was tested with a repeated single polarity (+) surge to ensure a worst case scenario.
21
www.bourns.com
6
The end of life shorting of the GDT is caused by the rapid breakdown of the emission coating and the
electrode material (metal) that further increases contaminates within the tube. The loose materials in the
tube attach themselves to the side of the ceramic body between the two electrodes. This has the effect of
causing a “virtual short” between the two electrodes.
Longevity and End of Life
Telcordia GR-974-CORE tests the primary protector to ensure adequate life of the tube in the field. The
2026-35 was subjected to 1000 impulse tests under a 500 A 10/1000 µs surge where the average DCBD
voltage stayed within its 20 % tolerance. GR-974-CORE specifies 100 repetitions at 300 A 10/1000 µs, so
this test setup can be considered well beyond normal requirements. The plan was to see what happened
at the actual end of life of the GDT. Figure 15 shows the results of the 2026-35 average DCBD voltage
and peak surge impulse voltage with 500 A 10/1000 µs over 1000 impulses. The blue boundary lines in
figure 15 highlight the minimum and maximum DCBD voltage limits specified as the tolerance in the
data sheet. It can be seen that the average DCBD voltage did not exceed this specification window in the
data sheet during 1000 surges.
Life Test with 500 A 10/1000 µs
900
800
700
Peak Voltage - V
6.1
600
500
400
300
200
0
200
400
600
Number of Surges
800
DC
Breakdown
1000
Surge
Impulse
Figure 15 – 2026-35 life test during 1000 surges of 500 A 10/1000 µs
Note: A good quality GDT reduces replacement in the field which ultimately saves cost. This is especially
important when units are deployed in remote locations and lightning rich areas.
22
6
GDT End of Life
900
850
800
750
Voltage
700
650
600
550
500
450
400
250
270
290
310
330
350
370
DC Breakdown Voltage
390
410
430
Surge
Voltage
450
Impulse
Voltage
Figure 16 – 2026-35 surge impulse vs DCBD voltage with 500 A 10/1000 µs
The surge impulse voltage from 500 A, 10/1000 µs (figure 16, orange dots) and the 1000 V/µs impulse
voltage (figure 16, yellow dots) was plotted against the DCBD voltage measurements over 1000 surges.
A direct relationship can be seen between the DCBD voltage and the surge impulse voltage under
repetitive surging. This ensures the equipment is still protected as the GDT comes to the end of its life.
Figure 14 shows that the DCBD voltage would have suddenly gone to a few volts highlighting that the
GDT had completely failed safe into a virtual short circuit.
6.2
The Dark Effect
GDTs have the reputation that the first surge on the tube will result in a higher breakdown than the next
successive surges. As the GDT is normally housed in a plastic module and deployed in a dark cabinet,
the term was called the dark effect. The initial strike ionizes the gas to make it settle into a consistent
breakdown voltage specification. First generation GDTs used to see the first breakdown voltage in the
region of 50 % to 70 % above the average sparkover rating. Equipment secondary protection would not
distinguish between the first and subsequent strikes, so this set the precedence on the voltage let-though
of the GDT. Today’s GDT still exhibits the dark effect, but Bourns has significantly reduced the impact
by changing the geometry and emission coating composition of the gas tube. Figure 17 shows the dark
effect with the 2026-35 series GDT with a 10 A 10/1000 µs. The average impulse let-through was
800 V, where the first surge impulse was only 11 % higher. Subsequent surges saw the impulse voltage
drop down 9 % (5 surges) from the average voltage.
Surge tests done with higher impulse current levels did not see this phenomenon, which indicates the
dark effect is dependent on the surge current. Higher surge currents also points to a higher rate of
voltage rise due to the fixed generator resistance, but further investigation is required.
23
www.bourns.com
6
2026-35 and the Dark Effect
1000
900
800
Peak Voltage
700
600
500
400
300
200
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20
Number of Surges
DC
Breakdown
1000V/µs
Impulse
10A 10/1000
Surge Impulse
Figure 17 – 2026-35 dark effect
Note: The Bourns® primary protectors such as the 2410 and 2420 have the dark effect included in the
tolerance of the GDT. The 5-pin 2410 and 2420 are tested to Telcordia GR-974-CORE to ensure that the
maximum peak voltage of the GDT is below 600 V under a 1000 V/µs rate of rise.
6.3
The Spark Effect
The 2026-35 was surged with 500 A 10/1000 µs to look at the dark effect at elevated impulse currents. At
500 A, the dark effect did not show in the test data results as the impulse voltage continued to increase
slightly under each subsequent surge test. There was a high increase in the DCBD voltage between the first
and second surges as shown in figure 18. This became known as the “spark effect”.
Note: The “Spark Effect” is a term created by the author and is not currently defined in any standards.
The spark effect is due to the arc being of a high enough energy density to cause contaminants (impurities)
to be released from the internal materials into the GDT gas atmosphere under a single surge. These
contaminants in the gas cause the increase of the DCBD voltage. Figure 18 shows an increase of 12 %
between the first two surges. Subsequent surges will trigger the “getter” effect of the special emission
coating that will attract the impurities (contaminants) and reduce the breakdown voltage to their original
level. Contaminants suspended in the gas will change the gas composition and decrease or increase the
breakdown voltage according to the Paschen curve of the particular gas mixture.
Figure 18 shows the DCBD voltage changing from 356 V to 398 V on the first surge (DCBD is measured
before the surge is applied). The 2026-xx series has a DCBD tolerance of ±20 % to give a range of 280 V to
420 V. The current failure tolerance point for Telcordia is 600 V (265 V for all surges up to 300 A 10/1000 µs)
where the 2026-35 is still well within specification.
24
6
2026-35 DC Breakdown Under 500 A 10/1000 µs
420
Peak Voltage (V)
400
380
360
340
320
300
0
5
10
15
20
Number of Surges
Figure 18 – Spark effect during repetitive 500 A 10/1000 µs surges
Note: The tolerance of the Bourns® GDT device includes the spark effect.
The second area of interest shown in figure 18 is that the DCBD reduces as the number of surges increase.
The emission coating on the inside of the GDT electrodes is activated due to the heat generated during
each strike. This has the effect of attracting the contaminants to the emission coating on the electrodes.
As the impurities reduce in the GDT gas atmosphere, the DCBD voltage also reduces. The DCBD voltage
and impulse surge voltage will stabilize. Figure 15 showed the DCBD voltage of the 2026-35 GDT settling
at 310 V after 447 strikes.
Note: The use of inferior electrode material will release more contaminates into the gas to cause premature
failure. Bourns® GDT products are designed for extended longevity.
6.4
GDT Resistance Below Its Breakdown Voltage
The resistance of a GDT and its capacitance (RC filter) indicates the insertion and return losses of the
component in high frequency applications. Bourns has independent test reports for insertion and return
loss of the 2410 and 2420 series of 5-pin modules up to 100 MHz. These reports are available on request.
Figure 19 shows the average resistance of the 2026-35 series GDT during 1000 surges of 500 A 10/1000 µs.
The resistance did not vary significantly during the test.
25
www.bourns.com
6
2026-35 Resistance with 500 A 10/1000 µs
3.5
3.45
Resistance (G-Ohms)
3.4
3.35
3.3
3.25
3.2
3.15
3.1
3.05
3
0
100
200
300
400
500
600
700
800
900
1000
Number of Surges
Figure 19 – Average resistance variation with 500 A 10/1000 µs surges
Note: Test measurement of such high impedances can be influenced by the equipment testing tolerance
and therefore any variation may not be a true representation of the GDT’s actual insulation resistance.
6.5
Section Summary and Recommendation
The Bourns® series of GDTs have been designed with longevity and reliability in mind where the 2026 series
can withstand a lot more repetitive surges than required by the current standards.
The old issue of the “dark effect” where the first strike would yield the highest surge impulse voltage has
been minimized in today’s product. Depending on the first impulse surge, a new effect called the “spark
effect” was seen. The increase in the surge impulse voltage is due to contaminates from the first strike
increasing the DCBD, which takes subsequent surges to rectify.
These voltage increases are within the DCBD tolerance of the Bourns® GDT to ensure quality, reliability
and dependability.
The Bourns® GDT will fail into a virtual short circuit where a reduction of the DCBD voltage is first
observed before the final rapid decline to a virtual short. There are occasions where a GDT can recover to a
level of if its DCBD, but a few subsequent surges will make it a permanent short.
26
7
Summary
A
GDT is often considered a slow operating overvoltage protector, but these devices are intended
to handle impulse currents many times higher than faster technologies such as solid state
(semiconductor) devices. The GDT has to develop enough voltage across the conductors to create an arc
across it. The gas between the contacts controls the electrical parameters and makes them repeatable under
a wide variety of environments. GDT technology has also evolved to provide “faster” switching devices that
reduce the impulse sparkover rating, but often at the expense of other key parameters such as tube life and
high glow to arc transition current or a high arc voltage. A GDT can also be designed with lower quality
materials such as the electrode material or electrode coating material for cost reduction, at the expense of
the GDT life in the field. Establishing the tradeoffs of the dynamic parameters, longevity and price is key to
designing in the correct product for the application. It might not be necessary to design in a GDT that will
last 20 years in a product that is expected to have only 5 years of service life.
A second area often questioned is that the electrical measurements of a GDT are never consistent. Some
of the experiments done in this document showed that this is indeed the case. The variation is due to
contaminants introduced into the gas from the normal operation of the GDT. Contaminants change the
electrical characteristics of the device where the emission coatings in the tube are designed to attract the
contaminants released into the gas. This results in the variations seen in the measurements during repeated
tests on the same device. Knowing these variations exist and having the manufacturer specify the tolerance
to take these variations into account allow the GDT to be designed correctly for a specific application.
Note: It is important to consider specifications for end of life tolerances, not just new tube tolerances.
This varies between GDT manufacturers.
Lightning strikes have been characterized as being a minimum of 30 kA up to 250 kA with a very short
duration, but the energy seen by a primary protector at the end of a twisted pair is more like a 500 A
10/250 µs surge. There are also GPRs (ground potential rises), that can have surges entering the ground
plane and equipment being damaged through the backplane. There is a belief that a faster rising voltage will
provide a higher impulse voltage. This is correct, but Bourns® GDT surge and ramp impulse voltage will
actually decline as the rate of rise of the surge voltages continue to increase as seen in figure 12. The ramp
or surge impulse voltage can now be calculated to find the expected peak voltage with a known DCBD
voltage option. This is simplified where the surge impulse voltage rating is just impacted by the shape of the
surge voltage waveform. It was observed that the 2026-xx series only requires a surge voltage of about 15 %
above the DC sparkover rating for a 10/1000 µs. This only increased to 50 % under a 2/10 µs type impulse
to ensure the GDT operates. This allows the maximum stress (GDT not operating) for the secondary
protection to be specified by just knowing the DCBD value of the GDT primary protector.
A common belief is to use the 1000 V/µs rating which is often 150 % above the DC sparkover rating and not
an accurate representation. Using the 1000 V/µs will provide a worst case scenario or provide maximum
tolerance, but at the expense of increased secondary protection requirements. A formula is now available
that will provide a more accurate method for defining coordination voltages to cover the wide voltage
options available in the GDT families.
Primary protectors have to meet Telcordia GR-974-CORE and GR-1361-CORE for the US and ITU-T K.12
for most other countries. Secondary protection needs to meet GR-1089-CORE or TIA-968-A for the US
and ITU-T K.20 and ITU-T K.21 for other countries. The design of primary and secondary protectors in
telecom systems has previously been totally exclusive of each other. Secondary protection standards such as
Telcordia GR-1089-CORE, issue 4 are the first step where the benefits of a GDT primary protector can now
be taken into consideration during the secondary protection design.
27
www.bourns.com
Appendix
8
Key Differences Between a GDT and TISP® Thyristor
Bourns also manufactures the semiconductor TISP® thyristor series that provides an overvoltage crowbar
protection solution similar to the GDT. A common question asked is “what are the differences between a
GDT and the solid state semiconductor TISP® thyristor”?
•
VDCBD vs repetitive peak off-state voltage, VDRM
Both define the maximum working voltage across the protector before conduction occurs where the
protector will have high impedance so that it will not interfere with the normal operation of the system.
The DCBD voltage is specified as a typical voltage where the tolerance has to be used to define the
minimum rating to not interfere with system’s operating voltage. The TISP® thyristor VDRM is specified
as an absolute maximum in the data sheets. The DCBD is measured by using a slow ramp voltage such
as a 100 V/s to 2000 V/s. The VDRM for the TISP® thyristor is measured at a specific current value and is
specified in the data sheet as a maximum of 5 µA at the VDRM value.
For example, a 2026-35, ±350 V DC sparkover with 20 % tolerance will have a minimum VDCBD of
±280 V. A TISP4350J3BJ has a minimum VDRM of 275 V.
•
Impulse sparkover voltage vs dynamic breakover voltage, V(BO)
Both define the maximum dynamic protection voltage window of the protectors. The protection voltage
is the maximum voltage the system will see. The GDT impulse voltage and the V(BO) impulse breakover
voltage of a TISP® thyristor are specified at a ramp voltage of 100 V/µs or 1000 V/µs.
For example, the 2026-35 has an impulse sparkover specification of 875 V at 1000 v/µs typical (about
200 % above the upper DC sparkover voltage rating) where the TISP4350J3BJ is specified with a maximum
of 364 V (32 % above the VDRM voltage rating). This shows the benefit of the TISP® thyristor with its tight
maximum working voltage to protection voltage (VDRM /V(BO)) window.
•
Impulse discharge current vs non-repetitive peak impulse current
These two parameters highlight the surge withstand rating of the protector where both are specified using
industry standard surge waveforms. The GDT will specify a short circuit current level and the minimum
quantity of operations it can withstand. The TISP® thyristor does not have a wear-out mechanism like the
GDT, so its impulse current ratings are specified as an absolute maximum.
For example, the 2026-35 can support a single surge of 40 kA 8/20 µs or 400 surges of 1000 A 10/1000 µs.
The TISP4350J3BJ is the largest current handling device in the DO-214AA (SMB) type package and is rated
for 800 A 8/20 µs and 200 A 10/1000 µs.
•
Capacitance
Capacitance of a GDT is significantly lower than a TISP® thyristor. The capacitance of a GDT is also not
affected by any bias or signal voltages across it and does not change with temperature. A TISP® thyristor
or any semiconductor overvoltage protector will have a capacitance value dependent on the surge rating
(which is dependent on silicon size) and will change according to the bias voltage across it. Thyristor
capacitance will also vary with the protection voltage of the same surge rating series where low VDRM
voltage options will have a higher capacitance value.
For example, the 2026-35 is specified with a capacitance of less than 2 pF. The 350 V TISP4350J3BJ
specifies a maximum of 105 pF with 2 V bias. Increasing the bias to 100 V will decrease the capacitance to
40 pF. A 70 V TISP4070J3BJ with the same surge rating as the TISP4350J3BJ has a maximum capacitance of
235 pF at 2 V bias.
28
9
References
Telcordia Technologies Generic Requirements, GR-1089-CORE, Issue 4, June 2006,
“Electromagnetic Compatibility and Electrical Safety –
Generic Criteria for Network Telecommunications Equipment”
Telcordia Technologies, GR-974-CORE, Issue 3, June 2002,
Generic Requirements for Telecommunications Line Protection Units (TLPU’s)
IEEE C62.31; IEEE Standard Rev. Dec. 2006,
Test Methods for low voltage gas tube surge protective device components.
ITU-T K.12, February 2006,
Characteristics of gas discharge tubes for the protection of telecommunications installations.
ITU-T K.20, July 2003,
Resistibility of telecommunication equipment installed in a telecommunications center to
overvoltages and overcurrents.
ITU-T K.21, July 2003,
Resistibility of telecommunication equipment installed in customer premises to overvoltages
and overcurrents.
TIA-968-A-4, November 2006,
Technical requirements for connection of terminal equipment to the telephone network.
29
www.bourns.com
My thanks go to Wolfgang Oertel, V.P of GDT Technology and Michael J. Maytum, Sr. Applications/
Standards Engineering Manager for critiquing the technical content of this document.
Disclaimer
Bourns assumes no liability for applications or customer product design. Customers are responsible for their products and
applications using Bourns® parts. To minimize the risks associated with customer products and applications, customers should
provide satisfactory design and working safeguards. It is the responsibility of the user to understand the safety practices relating to
working with electrical and electronic circuitry to avoid electric shock, fire hazards and explosions.
The information given here is based on data believed to be reliable, but because of the wide variety of situations in which this
information may be used, we cannot accept responsibility for loss or damage resulting from unsatisfactory performance. This
publication is not to be taken as license to use, or a recommendation to infringe, any patent. Information published by Bourns
about third-party products or services does not form a license from Bourns to use such products or services or a warranty or a
product endorsement.
Copyright
© 2008 Bourns with all rights reserved. Bourns must give written consent to reproduce any part of this publication. Personal
use of this material is permitted. The information given in this document is believed to be accurate and reliable, it does not form
part of any quotation or contract, and changes may occur without notice. Bourns will not accept liability for the results of its use.
Publication does not transfer nor imply any license under patent or other industrial or intellectual property rights.
30
Worldwide Sales Offices
Country Phone
Fax
Benelux:
Brazil:
China:
France:
Germany:
Ireland/UK:
Italy:
Japan:
Malaysia (KL Office):
Malaysia (Penang Office):
Singapore:
Switzerland:
Taiwan:
UK/Ireland:
USA:
+41 (0)41 768 5555
+55 11 5505 0601
+86 21 64821250
+33 (0)2 5473 5151
+49 (0)69 800 78212
+44 (0)1276 691087
+41 (0)41 768 5555
+81 49 269 3204
+60 3 71183138
+60 4 6581771
+65 6348 7227
+41 (0)41 768 5555
+886 2 25624117
+44 (0)1276 691087
+1-951-781-5500
+41 (0)41 768 5510
+55 11 5505 4370
+86 21 64821249
+33 (0)2 5473 5156
+49 (0)69 800 78299
+44 (0)1276 691088
+41 (0)41 768 5510
+81 49 269 3297
+60 3 71183139
+60 4 6582771
+65 6348 1272
+41 (0)41 768 5510
+886 2 25624116
+44 (0)1276 691088
+1-951-781-5006
Non-Listed European
Countries:
+41 (0)41 768 5555
+41 (0)41 768 5510
Region
Phone
Fax
Asia-Pacific:
Europe:
North America:
+886 2 25624117
+41 (0)41 768 5555
+1-951-781-5500
+886 2 25624116
+41 (0)41 768 5510
+1-951-781-5700
Technical Assistance
www.bourns.com
Bourns® products are available through an extensive global network of representatives and distributors.
To obtain technical applications assistance, a quotation, or to place an order, contact a Bourns representative in your area.
Specifications subject to change without notice. Actual performance in specific customer applications may differ due
to the influence of other variables. Customers should verify actual device performance in their specific applications.
Circuit Protection Solutions
"Bourns” and “MSP” are registered trademarks of Bourns, Inc. in the U.S. and other countries.
“TISP” is a trademark of Bourns, Ltd., and is Registered in U.S. Patent and Trademark Office.
COPYRIGHT© 2008, BOURNS, INC. • LITHO IN U.S.A. • 02/08 • e/GDT0719